Submission to

Productivity Commission:

Electricity Network Regulation Issues Paper


Submission to

Productivity Commission:

Electricity Network Regulation

Issues Paper

May 2012

Larry Kaufmann, Ph.D.

Senior Advisor

PACIFIC ECONOMICS GROUP

22 East Mifflin, Suite 302

Madison, WisconsinUSA 53703

608.257.1522 608.257.1540 Fax

1

Introduction

The Australian Government has asked the Productivity Commission (the Commission) to review the use of benchmarking as a means of achieving the efficient delivery of network services and electricity infrastructure. The Commission released an Issues Paper on this topic (as well as on the effectiveness of regulatory arrangements for interconnectors) in February 2012. The Issues Paper provides an introduction and overview of benchmarking in electricity regulation andpresents specific issues/questions on which it welcomes comment and formal submissions.

This submission presents my ownviews about the potential role of benchmarking in electricity network regulation. It also directs the Commission to a number of studies in Australia and elsewhere that pertain to benchmarking. There has been considerable amount of work on these issues in several jurisdictions that could be of value to the Commission’s inquiry, yet little of it was referenced in the Issues Paper.

I should note these comments reflect my work advising on benchmarking and related topics for nearly 20 years around the world. I have given expert witness testimony more than 30 times on benchmarking, total factor productivity (TFP) measurement, and the application of such metrics to energy utility regulation. I have also advised energy utilities and/or regulators on these issues in every State in Australia, as well as in New Zealand, the US, Canada, Japan, Germany, the UK, Mexico, Argentina, Bolivia, Jamaica, and Curacao. My assignment in Germany is perhaps particularly notable, since I was the leader of an international consortium advising the newly-created energy network regulator (the Bundesnetzagentur) on the worldwide experience with benchmarking in energy regulation and the lessons that could be gleaned from this experience and incorporated in Germany’s newly-established regulatory framework.

Relevant Research In Australia and Elsewhere

A significantamount of work has been undertaken on TFP measurement for electricity distributors in Victoria. I managed a series of projects on this issue, first on behalf of the electricity distribution businesses, and later on behalf of the Essential Services Commission of Victoria (ESC), over the period between 2001 and 2009. This work showed that it is currently feasible to measure TFP growth for Victoria’s power distributors accurately. Indeed, the empirical output of these projects certainly provides a strong foundation for a TFP-based approach to energy network regulation in Victoria, and perhaps in other States as well.

The main reports from these series of projects are the following:

  1. Incentive Regulation and External Performance Measures: Operationalising TFP – Practical Implementation Issues (June 2001)
  2. TFP Research for Victoria’s Power Distribution Industry (December 2004)
  3. Incentive Power and Regulatory Options in Victoria (May 2005)
  4. TFP Research for Victoria’s Power Distribution Industry: Update, (April 2006)
  5. TFP Research for Victoria’s Power Distribution Industry: 2005 Update, (November 2006)
  6. TFP Research for Victoria’s Power Distribution Industry: 2006 Update, (February 2008)
  7. TFP Research for Victoria’s Power Distribution Industry: 2007 Update, (December 2008)

If the Commission does not have copies of any of these papers, I would be happy to provide one.

In addition, on behalf of the ESC, I managed a research project estimating TFP growth for gas distributors in Victoria. I also worked co-operatively with ESC Staff to estimate TFP growth for a national sample of electricity distributors. I will not discuss these projects further since the Commission’s current inquiry pertains only to electricity rather than gas distribution, and because data constraints limited the quality of our preliminary estimate of TFP growth for Australia’s national power distribution industry.

Two of the papers cited above are primarily conceptual in nature. One of these is the first cited paper which, as the title suggests, was primarily a discussion of issues that would be involved with the practical implementation of a TFP-based approach rather than the “practical implementation” itself. This paper provided both an introduction and kind of a roadmap for the later empirical work that we undertook.

The second conceptual paper is Incentive Power and Regulatory Options in Victoria. In this report, we developed a complex but extremely flexible mathematical model which could be used to simulate how an average utility would respond to changes in its regulatory environment. This model was used to simulate the impact of different incentive regulation options on customer prices and utility profits under different regulatory approaches. One outcome of this report was that a “TFP-based” approach to regulation would generally yield better outcomes for consumers and companies than would various “building block” approaches to incentive-based, CPI-X regulation.

The other five papers cited above all develop estimates of TFP trends for Victoria’s electricity distributors and provide evidence on the regulated price trends that would have been generated for Victoria’s electricity distribution industry if a “TFP-based” approach had been used to set the business’s network prices. The most comprehensive of these reports is the first one prepared in December 2004, TFP Research for Victoria’s Power Distribution Industry. This report discusses our TFP methodology in detail, estimates TFP growth for each of Victoria’s electricity distributors and for the industry as a whole between the time of privatization in 1995 and 2003, and provides a host of supporting analyses. The subsequent reports update our estimates of industry TFP trends to include data for 2004, 2005, 2006, and 2007, respectively.

The outcome of this work is a rigorous estimate of electricity distribution TFP trends in the State that, over time, shows the emergence of a long-term TFP trend for the industry. To estimate this trend, however, it is necessary to measure the Victorian TFP trend from 1998 rather than 1995, since there was an identifiable, one-time “burst” of TFP growth between 1995 and 1998 (following privatization) which will not be repeated and is therefore not representative of the long-term trend. This fact was discussed extensively in PEG’s 2004 TFP report and has been evident in every reported TFP update in subsequent years.

Our work shows a clear trend emerging for electricity distribution TFP growth in Victoria. This is reflected in the graph below, which shows the annual average TFP growth for the Victorian electricity distribution industry, as this trend is updated annually for new information. The value for 2004 reflects average TFP growth for the industry from 1998 through 2004. The 2005 observation is equal to average TFP growth from 1998 to 2005. Similarly, the 2006 and 2007 observations are equal to the average growth in TFP for Victorian electricity distributors from 1998 through each of these respective years. This graph represents the actual “price path” that would result if PEG’s TFP study was, in fact, used in a TFP-based methodology, and PEG’s TFP index was updated annually to roll in new Victorian data.

Average TFP growth for Victorian electricity distributors was 1.24% over the 1998-2004 period, 1.07% over 1998-2005, 1.60% over 1998-2006, and 1.26% over 1998-2007. We believe this price path is relatively stable. It should also be noted that the volatility depicted above almost certainly exaggerates the volatility that would, in fact, result if PEG’s TFP specification was employed throughout Australia in a TFP-based approach. The reason is that the TFP growth trends plotted above correspond to average growth rates over six, seven, eight and nine year periods, respectively. TFP series almost always become more volatile as fewer years are used to compute the trend. I generally recommend that a minimum of nine years be used to compute a long-run industry TFP trend. Therefore, the series above reflects more volatility than would likely be experienced if PEG’s TFP specification was extended to all of Australia and ten or more years of Victorian data were initially used to compute this trend.[1]

Nearly all of this work was finalized at the time Victoria’s Department of Primary Industries (DPI) submitted a rule change application to the Australian Energy Market Commission (AEMC) to allow a TFP-based methodology to be used to set electricity distribution prices. This rule change application quickly led the AEMC to undertake a broad review into the use of TFP. In its Final Report, the AEMC concluded that a TFP-based regulatory option could create stronger performance incentives and improve the welfare of customers. However, instead of amending the rules to allow for the immediate implementation of a TFP-based regulatory option, the AEMC began a multi-year period for collecting the data that it believed would be necessary to implement a TFP-based approach. The rationale for this delay was that the AEMC believed it would be necessary for consistently-defined data to be gathered throughout Australia before accurate TFP measures could be calculated for the industry.

The AEMC’s review of TFP-related issues also led to a significant amount of debate between Economic Insights (EI) and myself, primarily regarding TFP measurement. Some of this debate drew on similar debates between EI and myself in two other jurisdictions: New Zealand and Ontario, Canada. The Commission has referenced one EI report from New Zealand, but for completeness it should also be aware of the following reports I prepared both in New Zealand and Ontario, as well as submissions I authored and which were presented during the AEMC review:

AEMC Review

  1. Essential Services Commission Submission to Review into the use of Total Factor Productivity for the determination of prices and revenues: Framework and Issues Paper (March 2009)
  2. Supplemental submission and spreadsheet-based models showing the impact on prices and earnings of TFP-based and building block approaches to CPI-X regulation (May 2009)
  3. Submission to Australian Energy Market Commission: Design Discussion Paper (October 2009)
  4. Submission to Australian Energy Market Commission: Preliminary Findings Report (April 2010)

New Zealand

  1. X Factor Recommendations for New Zealand Electricity Distribution Price Controls (July 2009)
  2. Reset of Default Price Path for Electricity Distribution Businesses: Submission to the Commerce Commission (August 2009)

Ontario

  1. Calibrating Rate Indexing Mechanisms for Third Generation Incentive Regulation in Ontario: Report to the Ontario Energy Board (February 2008)
  2. Defining, Measuring and Evaluating the Performance of Ontario Electricity Networks: A Concept Paper, Report to the Ontario Energy Board (April 2011)

Again, if the Commission does not have copies of any of these referenced papers, I would be happy to provide one. The last paper cited above could be particularly helpful to the Commission’s inquiry, since it is intended to be a reference document designed to help interested parties understand and evaluate the complexities that arise in discussions and applications of benchmarking and TFP measures.

TFP Measurement Issues: Resolved and Outstanding

There were four primary practical differences between the EI and myself regarding TFP specifications: 1) the use of physical or monetary metrics to measure capital input quantities; 2) the merits of adding unbilled outputs to the output quantity specification; 3) the use of revenues or marginal costs to weight output quantities; and 4) the complexity of the X factor formula and its ability to deal with firm-specific issues. This last issue is on the border between being ‘practical’ and ‘conceptual,’ but I will address it here since it does pertain to practical issues regarding TFP measurement.[2]

At the end of the AEMC review, three of these issues were effectively resolved in favor of the TFP specification I favored in the review and that I implemented in fact when estimating TFP for Victoria’s electricity distributors. These issues were the choice of the output specification (resolved to use “billed” outputs entirely); the choice of weights applied to these outputs when developing an overall output quantity index (resolved to use revenue shares rather than estimates of marginal costs); and the complexity of the X factor formula and its ability to deal with firm-specific issues (resolved that both approaches can in principle deal with firm-specific issues, but doing so requires X factors that are “tailored” to individual companies using either econometric methods or by estimating TFP for different groups of “peer” utilities which are subsets of the entire industry).[3]

One implication of the debates before the AEMC is that they show, for practical applications of TFP measures in utility network regulation, it is not true that “meaningful aggregate output measures are not necessarily easy to define,” as the Issues Paper states on p. 10. The Issues Paper cites an EI paper from New Zealand to support this claim, but EI itself chose not to rely on any of the analyses developed in that theoretical paper when it came time to make a practical choice for measuring output. The most meaningful, and appropriate, measures of aggregate output to be developed in regulatory applications of TFP are those employed in my TFP work for Victoria’s electricity distribution industry.

The one, remaining unresolved issue at the end of the AEMC review concerned the measurement of capital. EI continued to advocate physical metrics, while I continued to advocate deflated monetary values of capital expenditures, as the most appropriate measures for capital input. I welcome further inquiry by the Commission on this issue, for EI personnel are for all intents and purposes alone in their view on capital measures. For a comprehensive summary of the debate between EI and myself on this issue, the Commission can review Appendix Two of the Concept Paper I submitted in Ontario. I strongly encourage the Commission to evaluate this issue objectively and come to its own conclusions with respect to the best measure of capital inputs for benchmarking and TFP applications in regulation; a definitive analysis and statement by the Commission on how best to measure capital for regulatory applications of TFP and benchmarking would be a significant step forward in Australia.

Do Data Need to Be Improved before TFP-Based Regulation can Be Appliedin Australia?

My most fundamental concern with the AEMC Review was is its conclusion that a TFP-based regulatory option cannot be implemented for at least eight years. This conclusion was motivated by the need to develop a “robust and credible data-set” used to estimate industry TFP trends. Obviously, it is desirable to have high quality data, and it is acknowledged that data quality needs to be improved in Australia. But the AEMC’s conclusion that TFP-based regulation must essentially commence with eight years of fresh data is both unnecessary and undesirable, for a number of reasons.

One is that data quality is even more important for building block regulation, where regulated prices depend directly on the reported costs of individual companies. In building block regulation, data errors lead directly to price “errors.” This is not necessarily true in TFP-based regulation, where price changes depend on industry-wide changes in TFP and input prices. In spite of the data problems that currently exist, the AER is now using existing data to set prices under the building block methodology. Clearly, waiting for better data to become available is not an option for applying building blocks. If the current (imperfect) data are good enough to be used for setting regulated prices under the building block method, then these same data are good enough to use for calculating TFP trends. Indeed, since the regulatory consequences of using imperfect data are greater under building block than TFP-based regulation, data concerns actually argue for TFP-based regulation to be implemented more rather than less rapidly. Doing so reduces the potential for data “errors” to be directly reflected in regulatory prices.

In addition, it is not clear that current data will necessarily bias the computation of industry TFP trends. Indeed, industry TFP trends will not be biased by inconsistent or non-comparable data if those inconsistencies are random across utilities in the industry. Whenever this is true, data discrepancies or errors will tend to balance out across the cross section of firms, leaving the TFP index for the entire industry to be a good measure of the industry’s “real” index (i.e. the TFP index that would be measured using an internally consistent and comparable dataset across the industry). Moreover, the impact of data errors for any individual company to impact industry TFP is clearly diminished by the fact any individual company will be small relative to the industry.

In addition, for data errors to have a material impact on the TFP trend, they would have to impact the growth rate of TFP, not (in most instances) the level of the TFP index in any year. For example, if a data error in one year was entirely reversed in the following year, and both years were included in the sample period used to compute the TFP trend, the TFP trend would be unchanged. Even a one-time error in the industry (as opposed to individual company) data used to calculate TFP in any given year will have a smaller impact on the TFP trend, since flawed data from a single year will be averaged in with industry data from other years when computing the industry’s TFP growth rate over a multi-year period.